A secondary data centre is an investment in business continuity. But does every application need a fully running recovery environment?
AWS offers different recovery strategies, including pilot light and warm standby, so organisations can align infrastructure readiness with their recovery requirements. AWS Elastic Disaster Recovery (AWS DRS) supports a pilot-light approach by continuously replicating server data while keeping the recovery environment’s ongoing resource footprint relatively small. How AWS DRS replication works.
The right question is not simply, “Can we reduce our DR costs?” It is, “What recovery capability does each application actually need?”
Pilot light versus warm standby
These terms describe different levels of recovery readiness—not interchangeable names for a secondary site. Warm standby maintains a running, scaled-down application environment; pilot light requires additional resources or activation steps before it can serve application traffic.
| Consideration | Pilot light | Warm standby |
|---|---|---|
| Normal operating state | Core recovery resources and replicated data are maintained. | A scaled-down, functional application stack runs continuously. |
| During recovery | Launch or activate the remaining resources and validate the application. | Scale the existing environment to support production demand. |
| Readiness trade-off | More activation work before serving traffic. | Can already handle traffic at reduced capacity. |
Source: AWS Well-Architected disaster recovery guidance.
Rather than choosing one model for the entire organisation, start by asking application owners what downtime and data loss they can accept.
How AWS DRS works
AWS DRS can protect supported servers running on physical infrastructure, VMware vSphere, Microsoft Hyper-V and other cloud environments. Its replication agent first synchronises source disks, then continuously sends block-level changes into an AWS staging environment.
That staging environment includes Amazon EBS volumes, snapshots and EC2 replication servers. During a recovery event or drill, AWS DRS launches recovery instances on Amazon EC2 using the replicated data. AWS DRS core concepts.
Recovery targets need evidence
AWS describes DRS as enabling recovery point objectives measured in seconds and recovery time objectives measured in minutes. Treat those capabilities as a starting point for architecture and testing—not an unconditional promise for every business application. AWS Elastic Disaster Recovery overview.
Recovery point objective (RPO) expresses how much data loss is acceptable; recovery time objective (RTO) defines the acceptable interruption before service is restored. Recovery also involves validating the application and completing the additional steps documented in the DR runbook.
For an ERP platform, for example, make the test broader than “Did the servers start?” Ask whether users can sign in, complete a transaction and access the integrations they need.
Understand the complete cost
AWS DRS pricing includes a service charge for actively replicating source servers, alongside charges for the AWS resources used during replication, drills and recovery. Those resources include staging storage, snapshots and EC2 compute.
A meaningful estimate should therefore cover steady-state protection and the resources consumed when the recovery environment runs. AWS’s pricing examples explicitly separate ongoing replication costs from drill-related infrastructure costs. AWS DRS pricing and examples.
Before approving a design, request a cost model based on your server inventory, disk sizes, change rates, retention requirements and planned drill schedule—not just the DRS service fee.
Keep backup in the discussion
Pilot light and warm standby do not remove the need for backups: AWS recommends maintaining backups alongside replication in these recovery strategies. AWS DRS also supports recovery from previous points in time.
Your assessment should answer two separate questions: “How will we restart the service?” and “Which recoverable version of the data will we use?”
Ask your team to document retention requirements, recovery-point selection, restore validation and who authorises recovery. Make those decisions before an incident—not during one.
Test the recovery journey
AWS DRS supports recovery drills without disrupting the production environment or ongoing replication. Its recovery workflow still requires application validation, while activities such as traffic redirection and certain configuration changes sit outside the service. AWS recovery workflow guidance.
Use drills to test the complete operating procedure:
- Identify who declares a disaster and authorises failover.
- Validate application startup order and dependencies.
- Check user access and essential business transactions.
- Record actual recovery times against agreed objectives.
- Rehearse the return to the primary environment.
- Assign owners to every gap discovered.
The drill is complete when the business process works, not merely when the infrastructure starts.
Plan your next step
If you are reviewing a secondary data centre, planning an AWS migration or reassessing backup and recovery, begin with an application-level assessment rather than a product purchase.
Not sure whether pilot light or warm standby fits?
Talk to Cloud Is Good about an AWS migration, disaster recovery and backup assessment. Share your application inventory, current recovery setup and business priorities so we can scope the next step together.
Request an AWS Recovery Assessment →